Category-selective encryption protects chosen objects in UAV imagery
Original title: A Category-Selective Encryption Framework for Privacy Protection of UAV Images
An 80-class detector encrypts only user-specified categories, using thumbnail-preserving or XOR encryption.
The paper proposes a category-selective encryption framework for privacy in UAV aerial imagery. An 80-class detector identifies scene content, and only user-specified sensitive categories, such as pedestrians and vehicles, are encrypted, leaving other objects and background available for analysis. Two schemes are described: thumbnail-preserving encryption with adaptive pixelation, a learnable blend parameter, and recovery by a jointly trained Enhanced U-Net, and a parameter-free XOR stream cipher with exact reversibility. The work presents the architecture, composite loss, and joint training as a design foundation.
Unconfirmed: The paper presents the framework design and does not report a deployed surveillance system.
- Class-agnostic saliency methods obfuscate all salient regions, the authors argue.
- Thumbnail-preserving encryption recovers content with a jointly trained Enhanced U-Net.
- XOR stream cipher encryption is described as parameter-free and exactly reversible.
From the original article
1 min read at arxiv.orgImagery captured by unmanned aerial vehicle (UAV) systems increasingly raises privacy concerns, as UAV aerial surveillance routinely captures sensitive objects such as pedestrians and vehicles alongside ordinary scene content.
By Ting Yan · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.